+Miya y Nishida 2000 Filogenia Teleosteos

Molecular Phylogenetics and Evolution Vol. 17, No. 3, December, pp. 437– 455, 2000 doi:10.1006/mpev.2000.0839, available online at http://www.idealibrary.com on Use of Mitogenomic Information in Teleostean Molecular Phylogenetics: A Tree-Based Exploration under the Maximum-Parsimony Optimality Criterion Masaki Miya* ,1 and Mutsumi Nishida† *Department of Zoology, Natural History Museum & Institute, Chiba, 955-2 Aoba-cho, Chuo-ku, Chiba 260-8682, Japan; and †Ocean Research Institute, University of Tokyo, 1-15-1 Minamidai, Nakano-ku, Tokyo 164-8639, Japan Received January 13, 2000; revised August 1, 2000; published online November 30, 2000 INTRODUCTION We explored the phylogenetic utility and limits of the individual and concatenated mitochondrial genes for reconstructing the higher-level relationships of teleosts, using the complete (or nearly complete) mitochondrial DNA sequences of eight teleosts (including three newly determined sequences), whose relative phylogenetic positions were noncontroversial. Maximum-parsimony analyses of the nucleotide and amino acid sequences of 13 protein-coding genes from the above eight teleosts, plus two outgroups (bichir and shark), indicated that all of the individual protein-coding genes, with the exception of ND5, failed to recover the expected phylogeny, although unambiguously aligned sequences from 22 concatenated transfer RNA (tRNA) genes (stem regions only) recovered the expected phylogeny successfully with moderate statistical support. The phylogenetic performance of the 13 protein-coding genes in recovering the expected phylogeny was roughly classified into five groups, viz. very good (ND5, ND4, COIII, COI), good (COII, cyt b), medium (ND3, ND2), poor (ND1, ATPase 6), and very poor (ND4L, ND6, ATPase 8). Although the universality of this observation was unclear, analysis of successive concatenation of the 13 protein-coding genes in the same ranking order revealed that the combined data sets comprising nucleotide sequences from the several top-ranked protein-coding genes (no 3rd codon positions) plus the 22 concatenated tRNA genes (stem regions only) best recovered the expected phylogeny, with all internal branches being supported by bootstrap values >90%. We conclude that judicious choice of mitochondrial genes and appropriate data weighting, in conjunction with purposeful taxonomic sampling, are prerequisites for resolving higher-level relationships in teleosts under the maximum-parsimony optimality criterion. © 2000 Academic Press Key Words: mitogenomics; Teleostei; complete mtDNA sequence; long PCR; phylogenetic performance; taxonomic sampling; higher-level relationships. 1 To whom correspondence should be addressed. Fax: ϩ81-43-2662481. E-mail: [email protected]. Teleostei, the most diversified group of all vertebrates, comprises over 23,500 extant species (about 96% of all extant fishes) placed in 38 orders, 426 families, and 4064 genera (J. Nelson, 1994). They probably arose in the late Triassic, about 220 –200 million years ago (Carroll, 1988; but see Kumazawa et al., 1999), monophyly having been corroborated by many morphological characters (de Pinna, 1996). Some members of the basal Neopterygii, such as gars (family Lepisosteidae) and bowfin (Amia calva), and some fossil taxa have been thought to be, individually or in combination, candidates for a sister group of teleosts (Gardiner et al., 1996). Earlier classifications of the teleosts (e.g., Berg, 1940) were based on the recognition of a series of evolutionary grades, starting from a primitive group, through intermediate steps, toward the most advanced forms. As G. Nelson (1989) stated, such older classifications featured a bush at the bottom, comprising the lower teleosts (variously called Isospondyli or Clupeiformes), and a bush at the top of the tree, comprising the higher teleosts (Percomorpha). Following the publication of the seminal work by Greenwood et al. (1966) and the advent of cladistic theory (Hennig, 1966; Wiley, 1981), however, numerous comparative anatomical studies have been conducted in attempts to resolve both the basal (e.g., Rosen, 1973, 1974, 1985; Fink and Weitzman, 1982; Lauder and Liem, 1983; Fink, 1984; Begle, 1991, 1992; Johnson, 1992; Lecointre and Nelson, 1996; Johnson and Patterson, 1996) and the upper (e.g., Lauder and Liem, 1983; Stiassny, 1986; Stiassny and Moore, 1992; Johnson, 1993; Johnson and Patterson, 1993) “bushes.” Despite these efforts, there remains much controversy over higher-level relationships among the major teleostean lineages, being especially evident in the transitions of the classification systems adopted in the 1st to 3rd editions of J. Nelson’s “Fishes of the World” (J. Nelson, 1976, 1984, 1994). Molecular studies have been expected to prove deci1055-7903/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved. 437 although some incongruence (e.g. 1989. such as birds. Stepien and Kocher (1997) noted two obstacles to progress in the resolution of higher-level relationships. The first concerned the difficulty of alignment of highly divergent nucleotide sequences in hypervariable regions of ribosomal RNA (rRNA) genes. Gonostoma gracile (Miya and Nishida. although mitochondrial rDNA sequences are most powerful in resolving closely related species within a genus or a family (Bargeloni et al.. Indeed. Palumbi. Miya and Nishida. 1996. 1996). Meyer. 1997).438 MIYA AND NISHIDA sive in resolving persistent controversies over teleostean phylogeny (J. Additional complete mtDNA sequences have also been expected to be useful in answering some questions of fish relationships (Stepien and Kocher.. mitochondrial DNA (mtDNA) has many advantages as a phylogenetic marker because of its maternal inheritance and haploidy and the resulting short coalescence time (Avise.g. In mammals. 1999. and reported that the resulting maximum-parsimony tree agreed well with the morphologybased tree. 1997). no novel molecular phylogenetic hypotheses have been considered significant in recent studies dealing with higher-level relationships among major teleostean lineages (Lecointre and G. Using entire mitochondrial cytochrome b (cyt b) gene sequences from 31 fishes (including 30 teleosts). in which most exons are short (less than 500 bp. This is due apparently to the limited availability of polymerase chain reaction (PCR) primers for regions other than the cyt b and two rRNA (12S and 16S) genes.. although they have not yet fulfilled their promise (Stepien and Kocher. Hillis and Dixon (1991) had earlier suggested that nucleotide sequences from rRNA genes differing by more than 30% should be discarded as unalignable. With limited time and resources. they are unsuitable for resolving interfamilial or interordinal relationships in teleosts (Ortı ´ and Meyer. With the exception of the sister relationship of Clupeomorpha ϩ Ostariophysi (Le ˆ et al. see Kocher et al. what types of genes or genomes should we first explore for their usefulness in reconstructing higher-level relationships of teleosts? Although it remains unclear as to whether mitochondrial or nuclear genes are generally more efficacious for such a purpose.. 1991) suggested an unorthodox tree (e.. the transition from an apparently unsolvable to a solvable problem has come about mainly from the availability of complete mtDNA sequences (Penny et al. In that study. 1995. The second obstacle noted by Stepien and Kocher (1997) concerned “saturation” (see also Meyer. such weightings/translations also lower the amount of phylogenetic information. 1996). (1993) and Rubin and Dores (1995) analyzed amino acid sequences of growth hormone on the basis of different data sets. Moore. 1999). Ortı ´ and Meyer. Yamaguchi et al. 1994. overlapping segments of the entire genome . It appears that adequate resolution of higher-level relationships in any organism will require longer DNA sequences. (1993) criticized as “goofy” from the morphologist’s point of view. 1996. Springer et al. where nucleotide substitutions occur more frequently than those at the 1st and 2nd codon positions. Such a decrease in available phylogenetic information is significant. 1997. but also for partial sequences (300 – 600 bp in length) from individual mitochondrial proteincoding genes that have been routinely used in molecular systematics of fishes (usually cyt b gene. In addition. Groth and Barrowclough. Ortı ´ and Meyer (1997) observed that familial. fish-versatile primers that amplified contiguous. G.. We recently found that transfer RNA (tRNA) gene rearrangements have occurred in a teleost. Curole and Kocher. particularly those of the mitochondrial molecule. 1999). If so. 1996).. Le ˆ et al. comparing them with those from two outgroups. in which more than 60% of the sequences in the current database are from the cyt b gene (Sorenson et al. this is particularly notable at the 3rd codon positions in pro- tein-coding genes.. paraphyletic Neoteleostei) was observed. Reviewing the molecular systematics of fishes. Nelson. 1999). an earlier study using partial amino acid sequences from the three mitochondrial protein-coding genes (Normark et al. 2000). nonmonophyletic teleosts) that Patterson et al. 1994. 1998. however. Evidently. (1993) analyzed nuclear 28S rRNA sequences from 31 gnathostomes (including 18 teleosts) and found two highly supported nodes within the teleosts (Osteoglossomorpha ϩ Elopomorpha and Clupeomorpha ϩ Ostariophysi) that were incongruent with the morphology-based tree. 2000. 1997. not only for nuclear protein-coding genes. 1999). Bernardi et al.. including 25 and 24 teleosts. 1984. for instance. because most 3rd codon position substitutions do not cause amino acid replacements. it has been technically difficult to obtain a number of such longer mtDNA sequences from a wide variety of taxa. 1997). the entire mitochondrial genome was purified by gene amplification using long PCR and the products were subsequently used as templates for PCR with a set of 30 newly designed. 1993). 1989). see Table 1 in Lydeard and Roe. respectively.. In fact. 1997). 1994).or ordinallevel relationships among characiform and ostariophysan fishes could not be reconstructed with confidence using mitochondrial ribosomal DNA (rDNA) sequences. 1993. Lydeard and Roe (1997) reported that the resulting maximum-parsimony trees were largely congruent with the morphology-based tree. Nelson. Johnson and Patterson. Although this problem can be alleviated by simply excluding 3rd codon positions in protein-coding genes or by translating nucleotide sequences into amino acid sequences in phylogenetic analyses (Swofford et al.. 1999). however. A similar situation can be seen in other vertebrates. which have been sequenced more often (approximately 60% in total) than all other mitochondrial genes combined (Table 1. Nelson. but also allows accurate determination of complete mtDNA sequences that is faster than sequencing cloned mtDNA (Sorenson et al. (Miya and Nishida. their relative phylogenetic positions being noncontroversial from both the morphological and the paleontological points of view (Fig.. it should be useful for small. 12S and 16S ribosomal RNAs.8) 24 (0. 1999]). see Stiassny et al. we subdivided the three terminal branches of trout. X99772 [Johansen and Bakke. L29771 [Zardoya et al. ND1– 6. and determined their complete mtDNA sequences using the PCR-based approach with 98 fish-versatile primers. loach (Crossostoma lacustre. 1994]). Sorenson et al. unpublished data). 1999).0) 137 (3. Preliminary phylogenetic analyses using complete mtDNA sequences from the five species. cod). Diplophos taenia (Stomiiformes). with the exception of ND5. or endangered species (Miya and Nishida. U62532 [Noack et al. gracile. were relatively long and suggested that the unorthodox tree had resulted from the effects of “longbranch attraction” (Hendy and Penny. 1998]). 4L. Miya and M.5) 7 (0.1) a 12S and 16S rRNA. 1999. The major purposes of the present study were to explore the phylogenetic utility and limits of individual and concatenated mitochondrial genes for resolving higher-level relationships in teleosts and to provide a guideline for future molecular systematic studies of teleosts using mtDNA sequences. and cod with Coregonus lavaretus (Salmoniformes). ATPase subunits 6 and 8. Nishida. Scyliorhinus canicula. comprising nucleotide sequences from the 22 concatenated tRNA genes (stem regions only) plus several selected protein-coding genes (no 3rd codon positions) of phylogenetic performance appreciably better than the remainder. 1998). cytochrome c oxidase subunits I–III.. 1 in Waddell et al. Gonostoma gracile. plus two outgroups. NADH dehydrogenase subunits 1– 6. This taxonomic sampling scheme. 1999).. 1998). which also permits purposeful taxonomic sampling that increases phylogenetic accuracy (Hillis. We observed that all individual protein-coding genes.0) 49 (1. Miya and Nishida. D.5) 38 (0. we believe that we have overcome the technical difficulties in obtaining a number of longer mtDNA sequences from a wide variety of taxa.1) 602 (13.. and cod (Gadus morhua. 1995]). MATERIALS AND METHODS Taxonomic Sampling and Model Tree In addition to the complete (or nearly complete) mtDNA sequences from five teleosts. allowed us to evaluate the phylogenetic performance of the 22 concatenated tRNA genes and the 13 individual and concatenated protein-coding genes (both nucleotide and amino acid sequences) in comparison with the phylogeny expected from these eight teleosts.. which purportedly increases phylogenetic accuracy (strategy 4 in Hillis. lavaretus (Salmoniformes: Salmonidae). 1999) and further preliminary studies have indicated that this PCR-based approach for sequencing the complete mitochondrial genome is applicable to a wide variety of teleosts (M. Gonostoma gracile (AB016274 [Miya and Nishida. 1999).1) 5 (0. Accordingly.. M91245 [Tzeng et al. with all internal branches being supported by bootstrap values Ն90%.1) 3 (0. before further sequencing a number of teleosts. we determined the complete mtDNA sequences for three teleosts. carp (Cyprinus carpio. having no close relatives. 1989) owing to insufficient taxonomic sampling. 1999). 1999). However.4) 1172 (25. cytochrome b. ATPase 6 and 8.g.3) 139 (3. failed to recover the expected phylogeny of the eight teleosts.. best recovered the expected phylogeny. Also. taenia (Stomiiformes: presently unplaced. However. 1999.. Polypterus ornatipinnis. 1996). C. Kumazawa and Nishida.6) 693 (15. respectively.MITOGENOMICS AND TELEOSTS 439 TABLE 1 Approximate Representation of Mitochondrial Genes among 4577 Teleostean mtDNA Sequences in DDBJ (as of 8 August 1999) Genes a Cyt b Control region (D-loop) 12S rRNA 16S rRNA COI ND2 ATPase 6 and 8 ND4 and 4L ND6 ND3 COIII ND5 COII ND1 No. Fortunately. and Polymixia japonica (Polymixiiformes). the combined data sets.. G.2) 244 (5. cod being the most basal among teleosts) often received high statistical support irrespective of the weighting schemes or analytical methods employed (see also Fig. . (%) 1439 (31. and two outgroups (bichir. 1996. rare. The direct sequencings of the 30 PCR products were proven to be successful (Miya and Nishida. Thus. It should be noted that this approach not only greatly reduces the possibility of amplification of mitochondrial pseudogenes in the nuclear genome (Dowling et al. however. trout (Oncorhynchus mykiss. X61010 [Chang et al.2) 6 (0. the five existing complete (or nearly complete) mtDNA sequences are concentrated on the lower teleosts (up to the Paracanthopterygii). Cyt b. 1. we sought an efficient way to examine the phylogenetic performance of the various mitochondrial genes. 1996] and shark. revealed that unorthodox tree topologies (e. Y16067 [Delarbre et al. COI–III.. 1996]). 4L. Examination of the resultant trees indicated that the terminal branches of the three species (trout.. 1992]). G. taenia (CBM-ZF 10889 —western North Pacific off southern Japan). with modifications. No redundant taxa were used (e.g. lavaretus.e. 1993). Ostariophysi. japonica (Polymixiiformes: Polymixiidae). and other higher teleosts reflects the recently proposed hypothesis on the sister relationship of the former two groups on the basis of morphological and molecular evidence (see Lecointre and Nelson. The three species were purposefully chosen so as to subdivide the possible long branches of trout. Total genomic DNA was extracted using the Qiagen QIAamp tissue kit following the manufacturer’s protocol. The problematic Ateleopodidae were excluded from this figure (see Olney et al. Numbers of included taxa follow Nelson (1994).. gracile. This taxonomic sampling corresponds to strategy 4 in Hillis (1998). The relative phylogenetic positions of these eight teleosts are noncontroversial. and P. despite persistent controversies over teleostean phylogeny (see Stiassny et al.” and it was considered very good taxonomic sampling to increase phylogenetic accuracy (Hillis.440 MIYA AND NISHIDA FIG. 1996). Chiba (CBM-ZF): C. and cod. Natural History Museum and Institute. D. see Harold and Weitzman [1996]). Filled and open circles represent those species for which complete (or nearly complete) mtDNA sequences have been determined in the previous and the present studies. Saku Fisheries Experimental Station). i. 1. formerly in Gonostomatidae. Clupeomorpha ϭ Clupeiformes). Mitochondrial DNA Purification by Long PCR The mitochondrial genomes of the three species were amplified in their entirety using a long PCR technique . P. “select taxa within the monophyletic group of interest that are expected (based on current taxonomy or previous phylogenetic studies) to subdivide long branches in the initial tree. taenia...25 g) was excised from fresh specimens of each species and immediately preserved in 99. japonica (CBM-ZF 10890 —mouth of Tokyo Bay). D. Trichotomy among Clupeiformes. Names of the species for which complete mtDNA sequences have been determined in the present study are highlighted. Accordingly. lavaretus (CBM-ZF 10888 —from aquaculture pond. 0. (B) The expected phylogeny of the eight teleosts used in the present study.. we employed their relationships as a model tree in the present tree-based exploration of the phylogenetic performance of the mitochondrial genes. Fish Samples and DNA Extraction Mitochondrial DNA sequences were obtained from examples of C. 1996). A portion of the epaxial musculature (ca. 1998). Voucher specimens are deposited in the Fish Collection. and P. (A) Phylogeny of the major teleostean lineages according to Nelson (1994). which had been observed in the preliminary phylogenetic analyses (data not shown).5% ethanol. Note that none of these modifications affects the relative phylogenetic positions of the eight teleosts used in the present study. respectively. japonica. Zardoya et al.125 ␮l of 4 units Taq DNA polymerase (Z Taq..5 ␮l 10ϫ PCR buffer (TaKaRa). 1992. japonica. The primer pair amplifies nearly the entire.25 ␮l sterile distilled H 2O. see Mindell et al. were used for direct cycle sequencing with dye-labeled terminators (Applied Biosystems Inc.0 ␮l dNTP (4 mM).5 units LA Taq (TaKaRa). 2). cod. and 5 ␮l template. 1999). 1990]. the two primers may also be useful for other vertebrates with slight modifications (Fig. chicken. and AB034826 for P. carp. with two long PCR primer sequences (S-LA-16S-L and S-LA-16S-H) shown above the aligned sequences. . taenia. All sequencing reactions were performed according to the manufacturer’s instructions. Long PCR was done in a Perkin–Elmer Model 9700 thermal cycler and reactions were carried out with 30 cycles of a 25-␮l reaction volume containing 8. 38 such primers (Appendix 1) were newly designed with reference to the aligned. 1995.. 1985]. 2. with decreasing rates of variability at third. All DNA sequence electropherograms were carefully checked to determine whether they included the mitochondrial pseudogenes in the nuclear genome (for checkpoints. 1996a). Total genomic DNA was alternatively used as a PCR template for a region intervening between the two long PCR primers (S-LA-16S-L and S-LA-16S-H. AB034825 for D.6% SeaKem Gold agarose gel and later stained with ethidium bromide for band characterization via ultra violet transillumination. trout. with annealing and extension combined at the same temperature (68°C) for 16 –18 min. (1999). PCR was done in a Perkin–Elmer Model 9700 thermal cycler and reactions were carried out with 30 cycles of a 25-␮l reaction volume containing 14. 1981]).5 ␮l each primer (5 ␮M). Insertions/ deletions of specific nucleotides indicated by dashes (–). Two primers were designed in a highly conservative 3Ј end region of the 16S rRNA gene (S-LA-16S-L and S-LA-16S-H. Diplophos taenia (Dita).. TaKaRa) and 1 ␮l template (diluted long PCR products). lavaretus. Gonostoma gracile (Gogr).. Zardoya and Meyer. 2. Long PCR products were electrophoresed on a 0. purified by filtration through a Qiagen QIAquick column. Figs. 1994). 0. 2). Although originally designed for amplification of the teleostean mitochondrial genome.MITOGENOMICS AND TELEOSTS 441 FIG. Species names are abbreviated as follows: Coregonus lavaretus (Cola). 2.5 ␮l 10ϫ LA PCR buffer (TaKaRa).. Double-stranded DNA products from PCR. Tzeng et al. Aligned DNA sequences near the 3Ј end of the mitochondrial 16S rRNA gene from 10 fish species used in the present study. Johansen and Bakke. features that are consistent with mitochondrial origin were (1) presence of a conserved reading frame in protein-coding genes among all taxa.). 1994. 1996. 2. lungfish. PCR and Sequencing In addition to the 60 fish-versatile primers published in Miya and Nishida (1999). The thermal cycle profile was that of “shuttle PCR:” denaturation at 98°C for 10 s.5 ␮l each primer (5 ␮M).25 ␮l of 2.0% NuSieve agarose gel and stained with ethidium bromide for band characterization via ultra violet transillumination. 4. 2 and 3 and Appendix 1). Identity with the first sequence (shark) denoted by dots. and (3) no sequence changes yielding losses of known secondary structure for tRNA and rRNA genes. PCR products were electrophoresed on a 2. The contiguous PCR products overlapped each other by approximately 50 –300 bp. and extension at 72°C for 10 s. AB034824 for C. 0. 2. complete nucleotide sequences from the mitochondrial genome of six species of bony fishes (loach. 2. annealing at 52°C for 5 s.0 ␮l sterile distilled H 2O. The long PCR products were diluted with sterile TE buffer (1: 100) for subsequent use as PCR templates. (2) absence of extra stop codons.5 ␮l dNTP (4 mM). M10217 [Roe et al. As pointed out by Mindell et al. circular mitochondrial genome in a single reaction. and Polymixia japonica (Pxja). Chang et al. Noack et al.. 1996. so as to amplify the entire mitochondrial genome in a single reaction. Labeled fragments were analyzed on a Model 373S/377 DNA sequencer (Applied Biosystems Inc. frameshifts.. Primers used were the same as those for PCR. X52392 [Desjardins and Morais. The thermal cycle profile was as follows: denaturation at 98°C for 1 s. plus three other vertebrates (frog.. bichir. and second codon positions. or unusual amino acid substitutions. and human. The newly determined sequences are available from DDBJ/EMBL/GenBank databases under Accession Nos.). J01415 [Anderson et al. Fig. (Cheng et al. first. Diplophos taenia (B). control region. 3. ATPase 6 and 8. Also.) upon request. were excluded from the analyses. 12S and 16S ribosomal RNAs. 1993) for alignment of tRNA genes. 4L. tRNA loops and other ambiguous alignment regions. Transfer RNA genes are designated by single-letter amino acid codes. All protein-coding genes are encoded by the H-strand with the exception of ND6.. Gaps were considered as missing data rather than as fifth characters. 3. Individual gene sequence alignments for the eight teleosts and two outgroups were initiated with Clustal X (Thompson et al. ND1– 6. Relative primer positions are indicated with numerals designated in Appendix 1. such as the 5Ј and 3Ј ends of several proteincoding genes. leaving 916 and 11. Amino acids were used for alignments of the protein-coding genes and secondary structure models (Kumazawa and Nishida. Phylogenetic Analyses and Tree-Based Exploration Using Parsimony The DNA sequences were edited with the multiple sequence editor DNASIS (Hitachi Software Engineering Co. those encoded by the H-strand and L-strand being shown above and below the gene maps. 2000) were not feasible. such as preparing data matrices in NEXUS format. Aligned sequence data in NEXUS format are available from one of us (M. ND4L/ND4. cytochrome c oxidase subunits I–III. which is coded by the L-strand. cytochrome b. 1997) with default gap penalties and adjusted manually using SeqPup 0. MacClade ver. 4L. NADH dehydrogenase subunits 1– 6.6 (Gilbert. 1998. Since unambiguous alignments of two rRNA genes (12S and 16S) on the basis of secondary structure models (e. 12S and 16S. cyt b. they were not used in the analyses. and Polymixia japonica (C). Miya and Nishida.M. Ltd. Overlapping positions (two open reading frames) in several genes (ATPase 8/ATPase 6.442 MIYA AND NISHIDA FIG.g. exporting tree files. and exploring alternative tree topologies. 1993). ATPase 6/COIII. ATPase subunits 6 and 8. respectively.08 (Maddison and Maddison. Branch-and-bound maximum-parsimony (MP) analyses for amino acids and nucleotides were conducted with 500 bootstrap replicate searches using PAUP . Gene organization and sequencing strategy for the mitochondrial genomes of Coregonus lavaretus (A). ND5/ND6) were duplicated in the analyses. COI–III.. to circumvent those longer than one or two bases being taken as representing multiple events (Swofford.151 nucleotide positions from the 22 tRNA and 13 protein-coding genes. respectively.). 1992) was used in various phases of the phylogenetic analyses. CR. 1996). respectively) (Fig. All phylogenetically uninformative sites were ignored. 4. was identical to that of typical vertebrates. lavaretus). Concatenated tRNA Genes Kumazawa and Nishida (1993) demonstrated that nucleotide sequences from the mitochondrial tRNA- FIG. Since there was no a priori reason to select a specific model of sequence evolution for such a tree-based approach. retention index [RI] ϭ 0. Since the three gene lengths were very similar (Appendix 2). The single most-parsimonious tree for eight teleost species and outgroup taxa. which were encoded on the L-strand (Appendix 2). including that of another stomiiform (D.418 bp (D. 1988. Phylogenetic performances of the individual and concatenated protein-coding genes were evaluated by calculating the differences in the number of steps between the unconstrained most-parsimonious tree and the constrained most-parsimonious tree (ϭexpected phylogeny). 1999) and D. and (4) using amino acid sequences (AA).737 bp (C. not least because of their slow evolutionary rate. with a tree length of 728 steps (consistency index [CI] ϭ 0. In addition. Phylogenetic analyses were conducted separately and in combination. taenia) to 16. the 22 tRNA-stem region sequences were concatenated (total 916 bp) and subjected to unweighted MP analysis. except for the ND6 and 8 tRNA genes. gracile (Miya and Nishida. as have been recently found in the tRNA Glu–tRNA Pro region for a stomiiform. Since the individual tRNA genes were too short (66 –75 bp) to analyze separately. For the protein-coding genes. the differences were statistically evaluated by comparing the distribution of character state changes per site using Wilcoxon signed-ranks tests (Templeton. 1995). gracile is an autapomorphic feature revealed by limited taxonomic sampling. Furthermore. 1996) and lamprey (Lee and Kocher. using 22 concatenated tRNA genes (stem regions only) and 13 individual and concatenated protein-coding gene amino acids and nucleic acids. although a slight increase and a marked decline in bootstrap values were observed for Clades B (88%) and C (56%). as found in the bichir (Noack et al. 1999). because the tRNA Ser(AGY) genes in G. 1999). who considered those Ն95% as significant. the differences were due mostly to length variations in the control region (765–1076 bp). and rescaled consistency index [RC] ϭ 0. 22 tRNA. 1996b). total. and 13 protein-coding genes. two other internal branches (Clades B and C) were supported by moderate bootstrap values (82 and 75%. the genome organization of the three species. However. but see Kumazawa et al. 1983) implemented in PAUP 4. (3) excluding 3rd positions (NC [no 3rd]). it was evident that the tRNA-stem sequences alone were able to reproduce higher-level relationships of teleosts with divergence times dating back to at least the upper Cretaceous..248). our analyses relied predominantly on cladistic parsimony. analyses were conducted at four levels: (1) using all nucleotides (NC [all]). gracile (Miya and Nishida.456. as found in other vertebrates (Appendix 2).0b2a. approximately 65–95 million years ago (Carroll. and the differences were subsequently standardized for the length of each gene by dividing those values by the number of nucleotides/amino acids in each gene (Zardoya and Meyer. taenia).. taenia had no recognizable DHU stem and loop.0b2a (Swofford. most genes were encoded on the H-strand. derived from 22 tRNA-stem region sequences. Although the present bootstrap values for the tRNA-stem phylogeny did not meet the criterion of Kumazawa and Nishida (1993).544. . G. The genome content of the three species included 2 rRNA. stem regions were useful for resolving deep-branch phylogenies among major vertebrate lineages (20 – 600 million years of divergence). Bootstrap values were very high for contemporary sister species’ pairs (Clades D–G. The resulting tree topology was identical to that of the expected phylogeny (Fig. The total lengths of the mitochondrial genomes in the three species ranged from 16. Inclusion of other unambiguously aligned sequences from the tRNA-loop regions (502 bp. RESULTS AND DISCUSSION Genome Organization Our initial expectation was that some of the newly determined complete mtDNA sequences from the three species would exhibit gene rearrangements. respectively. indicating that the characteristic gene order of G. 4). (2) excluding 3rd position transitions (NC [no 3rd TS]). 95–100%) and for a clade that corresponded to Euteleostei (Clade A. Numbers above branches indicate bootstrap support obtained for 500 replicates using branch-and-bound algorithm. 1418 bp) did not alter the tree topology. plus the control region. 100%). As in other vertebrates.MITOGENOMICS AND TELEOSTS 443 4. 1998). they were excluded from the analysis. 4). only the rooting of such leading to confusion (Curole and Kocher. the most-parsimonious tree (tree length ϭ 9564 steps.166). When the 3rd codon position TSs (NC [no 3rd TS]) were excluded from the data set.349. and (D) amino acids. 1996a. 1998).001). G. or “bizarre” (Cao et al.. It should be noted. 5A). By excluding all of the 3rd codon positions (NC [no 3rd]) from the data set. gracile]) being supported by a bootstrap value of 100% (tree length ϭ 15. CI ϭ 0. MP analysis still failed to recover the expected phylogeny (Fig. 5. Unweighted MP analysis using all nucleotides (NC [all]) failed to yield the expected phylogeny (Fig. although other investigators have variously interpreted unorthodox tree topologies of major vertebrate lineages resulting from their own studies as “odd” (Zardoya and Meyer. These investigators have regarded such hypotheses as novelties that challenged previous. on the basis of long-standing morphological evidence. Numbers above branches indicate bootstrap support obtained for 500 replicates using branch-and-bound algorithm. the finding that divergence between Amniota and Osteichthyes predated divergences among lungfish and all other bony fishes (Rasmussen et al.b). taenia. (C) nucleotides excluding entire 3rd codon positions (NC [no 3rd]). In any event. including a sister relationship for marsupials and monotremes. CI ϭ 0. C. The strict consensus of the most-parsimonious tree for eight teleost species and outgroup taxa.359. Moreover. that all of these data sets arrived at the same basic tree topology of gnathostomes. 1997). 5B).. well-established phylogenies. RC ϭ 0.530. derived from the 13 combined protein-coding genes. “incorrect” (Naylor and Brown. P ϭ 0. RI ϭ 0. lavaretus] ϩ [D. these results indicate explicitly that use of combined protein-coding gene sequences does not always support the expected. indicating that the unweighted use of all codon positions positively misled the analysis.891. P ϭ 0. The present study also demonstrated that use of all of the 13 protein-coding genes did not always recover the expected teleost phylogeny. exclusive of placental mammals (Janke et al.313. 1998). RC ϭ 0. however. Combined Protein-Coding Genes Recent phylogenetic analyses of the combined 12 (excluding ND6 owing to its heterogeneous base composition) or 13 mitochondrial protein-coding genes for vertebrates have resulted in several controversial hypotheses (Mindell et al.444 MIYA AND NISHIDA FIG. RI ϭ 0. 1999). however..170) was statistically indistinguishable from the expected phylogeny (z ϭ Ϫ0. 1998). 1996. using (A) all nucleotides (NC [all]). 1997. tree length difference between the unconstrained most-parsimonious tree and the constrained most-parsimonious tree (ϭexpected phylogeny) (54 steps) was highly significant (z ϭ Ϫ3..474. well-established phylogenies. with an incorrect branch ([trout. (B) nucleotides excluding 3rd codon position transitions (NC [no 3rd TS]).373).690 steps. 1999). MP analysis barely . 1999). and that cartilaginous fishes have a terminal position in the piscine tree (Rasmussen and Arnason. 1533 steps (NC [no 3rd TS]. Unweighted MP analysis of the amino acid sequences (AA). Cummings et al.159).. although some deep branches. At the amino acid level. such as Clade C. RI ϭ 0. ND5. 5D). 1995. ATPase 6 in NC (no 3rd TS). Zardoya and Meyer. since the unconstrained most-parsimonious trees were not significantly different from the constrained most-parsimoni- . Individual Protein-Coding Genes Recent studies have demonstrated that individual mitochondrial protein-coding gene sequences were insufficient in themselves for inferring the phylogeny of the taxa examined (Cao et al. Fig. all positively misled the analyses (Fig.561. characterized by good phylogenetic performances (Fig. suggested that most mitochondrial protein-coding genes had the potential to recover the expected phylogeny. 7). and 723 steps (NC [no 3rd]. and ND4L and ND2 in NC (no 3rd). were consistently supported by only low bootstrap values (39 –58%). In that case. avoidable exclusion treatment could have effectively eliminated misleading. 1992) (data not shown).169). no gene was able to reproduce the expected phylogeny. 1998. The single most-parsimonious tree for eight teleost species and outgroup taxa. 7). whereas ND4.352. It was apparent that not only the 3rd position TS. the only protein-coding gene that consistently reproduced the expected phylogeny was the ND5 gene (Fig. Successive Concatenation Analyses Only the ND5 gene consistently reproduced the expected phylogeny.553. and RC ϭ0.365. although Naylor and Brown (1997. 5C) resulted. It is likely that such un- FIG. successfully recovered the expected phylogeny (tree length ϭ 2792 steps.216. only one of the three equally most-parsimonious trees was congruent with the expected phylogeny (data not shown). Fig. The other 12 protein-coding genes were unable to reproduce the expected phylogeny. 6. characterized by noticeably poor phylogenetic performances. 1999) did not improve the tree statistics or bootstrap values (data not shown). probably owing to insufficient phylogenetic information being contained within a single gene. 1996. which required more nucleotide changes. however. and ND4L always occupied the worst three genes. with high bootstrap support (97–100%) for all of the internal branches except two (61 and 74%. 6).313. ND6 and COI in NC (all). Mindell et al. irrespective of the use of nucleotide sequences with various weighting schemes or the use of amino acid sequences..202). with a tree length of 2464 steps (NC [all].. RI ϭ 0. 1998) suggested that isoleucine (I). CI ϭ 0. CI ϭ 0. Graybeal. ND6. fully bifurcating most-parsimonious tree. RC ϭ 0. that relatively long sequences at the 5Ј and 3Ј ends of the ND5 gene (total 63 positions) had to be excluded from the analyses owing to ambiguous alignments in those regions. CI ϭ 0. COIII. The phylogenetic performances of each of the 13 protein-coding genes were compared on the basis of the standardized tree length difference between the mostparsimonious tree and the expected phylogeny (Fig. 5D).386. and RC ϭ 0. Although there were no statistically significant differences for most genes in most analyses. It should be noted that ATPase 8. The above comparisons of phylogenetic performance. 1994. however. exclusion of these three amino acids by converting L and V into I from the data matrix (see Cao et al. CI ϭ 0. and valine (V) were the amino acids responsible for phylogenetic inconsistency.MITOGENOMICS AND TELEOSTS 445 recovered the expected phylogeny. CO I. with the exception of the COIII gene when the 3rd codon positions were entirely excluded. Numbers above branches indicate bootstrap support obtained for 500 replicates for each weighting scheme (NC [all]/NC [no 3rd TS]/NC [no 3rd]) using branch-and-bound algorithm. by using the PROTPARS weight matrix for amino acids provided in MacClade (Maddison and Maddison. CI ϭ 0.481. 1994. although three equally most-parsimonious trees (tree length ϭ 4009 steps. however. 1999. on the other hand. homoplasious signals from the data set. RI ϭ 0. 1996b)... Our analyses also indicated that most of the genes consistently failed to yield the expected phylogeny. At the nucleic acid level. under any of the weighting schemes. and cyt b always appeared in the best six genes. but also the 3rd position TV differences contained much homoplasy that prevented the recovery of the expected phylogeny.305. RI ϭ 0. There was no noticeable improvement in tree statistics or bootstrap values when greater weight was given to amino acid replacements. 7).697. MP analyses of the ND5 gene sequences using the branch-and-bound algorithm for the three weighting schemes produced a single.449. Bootstrap supports for the internal branches were generally high for relatively shallow branches in all weighting schemes.523. leucine (L). Similarly. RC ϭ 0. ND6. Fig. Russo et al. and RC ϭ 0. ND1. RI ϭ 0. derived from the ND5 nucleotide sequences. It should be mentioned. cytochrome c oxidase subunits I–III. 8A). 4L. was not reproduced at all. 7) and can then expect to find an increasingly more robust phylogeny as more data are collected. 1998). Of course. Differences were standardized for each gene length. ultimately resulting in the successful recovery of the expected phylogeny. 8B). 8A). Statistically significant differences (P Ͻ 0. Fig. in which genes are successively concatenated in order of good phylogenetic performance (Fig. an investigator is lucky enough to find the best gene at first (see Fig. ous trees (ϭexpected phylogeny. It should also be noted that the likelihood of . such as the level of the analyses (nucleotides or amino acids). 7). then concatenation of genes should increase such information effectively. 4L. individual protein-coding genes were concatenated one after another in the following two ways. as has been demonstrated in the above analyses of the combined protein-coding genes (see Fig. in other words. By comparing changes in the phylogenetic performance of the concatenated genes under these two different conditions. even if all 13 protein-coding genes were included in NC (all) and NC (no 3rd TS) analyses (Fig. 8): (1) the worst possible case. 7) and attempts to sequence other genes one after another in the hope that misleading signals will eventually disappear as more data are collected. 7). it appears that such a simple “more data” approach does not warrant successful recovery of the expected phylogeny.05) are indicated by asterisks. methods of concatenation.446 MIYA AND NISHIDA FIG. Phylogenetic performance of the mitochondrial protein-coding genes as evaluated by differences in step number between the unconstrained most-parsimonious tree and the constrained most-parsimonious tree (ϭexpected phylogeny). and were subsequently subjected to MP analyses so as to evaluate any changes in phylogenetic performance of the concatenated genes (Fig. and (2) the possible best case. 5). and weighting schemes for nucleotide sequences. cyt b. in which genes are successively concatenated in order of poor phylogenetic performance (Fig. however. The assumptions implicit in this approach are that any incorrect phylogenetic inferences will be due to stochastic error associated with an insufficiently large sample size of characters and that they will disappear as more data are collected (Naylor and Brown. it was evident that the standardized difference between the unconstrained most-parsimonious tree and the constrained expected phylogeny decreased in all analyses as more data were collected. 7. ND1– 6. Standardized values were sorted to indicate performance ranking. The expected phylogeny. COI–III. NADH dehydrogenase subunits 1– 6. Arrows indicate that the unconstrained most-parsimonious tree was identical to the expected phylogeny. To evaluate the effects of the gene concatenation on the phylogenetic analysis. an unlucky inves- tigator begins sequencing from genes with poor phylogenetic performance (see Fig. cytochrome b. If the unsuccessful recovery of the expected phylogeny was simply due to insufficient phylogenetic information in a single gene. When genes were concatenated in order of poor phylogenetic performance (Fig. 8A). ATPase subunits 6 and 8. ATP6 and 8. we expected to gain an insight into the preferred methods of data handling. on the basis of the phylogenetic performances (Fig. in other words. . which were standardized for the number of phylogenetically informative positions for each gene. respectively. on the other hand.0600 to Ͻ0. were concatenated.0001). 8. Arrows indicate that the unconstrained most-parsimonious tree was identical to the expected phylogeny. Statistically significant differences indicated by asterisks (*0. recovering the expected phylogeny noticeably decreased as more data were collected in the former analysis (P ϭ 0. **0. indicating that historical signals overcame homoplasious signals as more data were collected. even if genes with poor phylogenetic performance were initially included in the analyses.MITOGENOMICS AND TELEOSTS 447 FIG. Phylogenetic performance was evaluated by differences in step numbers between the unconstrained most-parsimonious tree and the constrained most-parsimonious tree (ϭexpected phylogeny).01.01 Ͼ P). In the NC (no 3rd) and AA analyses (Fig. 8A). the expected phylogeny was recovered when 13 and 10 or more protein-coding genes.05% Ͼ P Ն 0. Changes in the phylogenetic performance of successively concatenated protein-coding genes in order of (A) good performance and (B) poor performance. in which the 13 protein-coding genes were classified into three groups of good (ND4.0015 0. the important question lies in how to discriminate phylogenetically “good” genes from the 13 protein-coding genes.0000 0. 8A and 8B). 8A and 8B). ND5. ND2). no genes with “poor” phylogenetic performance in Zardoya and Meyer (1996b) appear in the “very good” and “good” categories in this study.0337 0. the above analyses also indicated that. statistical differences between the most-parsimonious tree and the expected phylogeny became more significant as more data were collected (Fig. ND1.0127 0. COIII.0000 0. and cyt b were the best genes for reconstructing higher-level relationships among major vertebrate lineages. and poor (ATPase 6. If this is so. cyt b.0308 0.0037 0.1091 Total 0.0036 0. the 13 protein-coding genes could be roughly classified into the five groups of very good (ND5. ND4. ND6.0223 0.0177 0.0020 0.0117 0. c Nucleotide sequences excluding 3rd codon positions. Although we have no a priori knowledge regarding the phylogenetic performance of mitochondrial protein-coding genes for a specific systematic question. In the former analysis.0038 0.0106 0. a different picture emerged.0057 0. otherwise. COIII. the inclusion of genes with poor phylogenetic performance prevented the recovery of the expected phylogeny in these analyses. Interestingly. 7).0405 0. 8B).0087 0. Evidently. in which the phylogenetic performance of the ND5 gene was far better than that of any others. ND4L) genes. the expected phylogeny was consistently reproduced through successive concatenation in the NC (no 3rd) and AA analyses. COI).0264 0. In contrast. several genes with better phylogenetic performance should be chosen in the analyses. medium (COII. d Amino acid sequences. Although somewhat arbitrary. Also. being standardized for each gene length (see Fig. (1996) recommended that ND5. and none with “very good” and “good” phylogenetic performance in this study appear in the “poor” category in Zardoya and Meyer (1996b) (Table 2).0036 0.0135 0. It should be noted that the top six genes (classified as . poor (ND1.0073 0. Furthermore.0172 0. however.0022 0. b For abbreviations of genes.0087 0. 8B). ND2. homoplasious signals as more data were collected in the NC (all) and NC (no 3rd TS) analyses.0540 0. and very poor (ND4L.0038 0.0204 0. ND4.0174 0.0229 0. Russo et al. ATPase 6).0040 0. medium (ND3. ATPase 8. COI).1576 Present study Very good Very good Very good Very good Good Good Medium Medium Poor Poor Very poor Very poor Very poor Zardoya and Meyer (1996) Good Good Medium Good Medium Good Poor Good Medium Poor Poor Medium Poor a a Differences in number of steps between the unconstrained most-parsimonious tree and the constrained most-parsimonious tree (ϭexpected phylogeny).0071 0. historical signals would not become apparent until all 13 protein-coding genes were combined (see Figs. in the latter. it was evident that historical signals were overwhelmed by stronger. Discriminating “Good” Genes As the above two analyses explicitly demonstrated (Figs. good (COII. at least three or more genes were required to reproduce the expected phylogeny.0020 0.448 MIYA AND NISHIDA TABLE 2 Phylogenetic Performance of the 13 Mitochondrial Protein-Coding Genes Phylogenetic performance a Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 Genes b ND5 ND4 COIII COI COII Cyt b ND3 ND2 ND1 ATP6 ND4L ND6 ATP8 NC (no 3rd) c 0. When genes were concatenated in order of good phylogenetic performance (Fig.0085 0. cyt b). it might be of some help to discriminate “good” genes on the basis of the phylogenetic performance from the two data sets (NC [no 3rd] and AA) that successfully recovered the expected phylogeny (Table 2).0510 0.0485 AA d 0. ND6). ND3. ATPase 8) genes on the basis of the total score of phylogenetic performance from the two data sets (Table 2).0306 0.0244 0. Furthermore. see footnote of Table 1. At the nucleic-acid level. even if an investigator uses such data sets. this classification of the phylogenetic performance roughly corresponds to that noted in Zardoya and Meyer (1996b) for reconstructing higher-level relationships among major vertebrate and mammalian lineages (Table 2). a simple “more data” approach was shown to be successful only when either NC sequences with an appropriate weighting (NC [no 3rd]) or AA sequences were used in the analyses. it is notable that all of the genes with “very good” or “good” phylogenetic performance in this study have been more or less considered “good” genes for reconstructing higherlevel relationships of vertebrates in one or more studies (Zardoya and Meyer. CONCLUSIONS The present study embodied the earlier expectation of Kumazawa and Nishida (1993) that the use of mitochondrial tRNA genes in combination with appropriate parts of the mitochondrial protein-coding genes may become a feasible strategy. Moreover. ND4. COI. and Russo et al. Nevertheless. In all three analyses. 1996b. that the universality of these findings are unclear and that more empirical studies. Subsequent inclusion of genes with “medium” or poorer phylogenetic performance into the data sets generally lowered bootstrap values. (1996). in combination with the 22 concatenated tRNA genes. Changes in bootstrap values for these two clades were similar in different data sets. and cyt b. 9C): all internal branches were supported by bootstrap values Ͼ90% when 3 to 5 selected protein-coding genes were used in combination with the 22 concatenated tRNA genes. 8B). Since the expected phylogeny was recovered in all data sets (except that comprising only the ND5 gene in AA analysis. and COIII) and cytochrome b (the amino acid sequences of all of these being highly conservative. 1995). It appears that their relatively poor phylogenetic performance is due partly to gene lengths. it should be emphasized that purposeful taxonomic sampling (Hillis. To the best of our knowledge. 1996. Bootstrap support for two other clades (Clades B and C) varied considerably. from 62 to 100% for Clade B and from 22 to 95% for Clade C.” “poor. we believe that more accurate phylogenetic estimation is feasible by incorporating knowledge of the molecular structures and processes into the phylogenetic inference (Naylor and Brown. should be conducted for other animals at various taxonomic levels. with the exception of data sets comprising the ND5 gene alone (Fig. 9). Indeed. the expected phylogeny was not recovered without appropriate weightings and sufficient numbers of characters (see Figs. 9). the phylogenetic performance of the concatenated genes was evaluated by the level of bootstrap support for each internal branch (Fig. Nevertheless. Because of the numerous combinations of the 13 protein-coding genes. at the same time imparting greater variability. Also.. It was also evident that not only the 3rd codon position TSs. Instead of enumerating all possible combinations. 1998). the remaining five. however. probably because genes yielding correct results might vary among data sets and thus not be determinable a priori (Naylor and Brown. without exception. and ND6) and two ATPase subunits (ATPase 6 and ATPase 8) were included in the “medium. as has been pointed out by Cummings et al. Zardoya et al. 9).MITOGENOMICS AND TELEOSTS 449 “very good” or “good”) in this study comprised three cytochrome c oxidase subunits (COI. ND2. ND4L. with 100% bootstrap support for all but three nodes. We acknowledge. Naylor and Brown. 5A. Furthermore. In this study.” and “very poor” categories. ND3. (1995.. 5B. Naylor and Brown (1998) reported that the expected chordate phylogeny was recovered from a combined data set comprising amino acid sequences from ND1. but also the 3rd codon . and possibly to evolutionary rate differences between these genes (Zardoya and Meyer. in the hope that stronger historical signals in the tRNA data (judging from the tree statistics) would strengthen the statistical support for the internal branches. Preferred Data Set Zardoya and Meyer (1996b) have suggested that the use of a combination of genes with better phylogenetic performance is likely to lead to increased accuracy in phylogenetic reconstruction. it is impractical to attempt to chose the best combination of genes (how many genes and which?) to recover the expected phylogeny. Although no “appropriate” protein-coding genes have been explicitly proposed since then. Russo et al. the bootstrap support values for Clades A and D–G were consistently Ն95%. historical signals embedded in these sequences are strengthened by combining the two different data sets. however. 1996b). the use of some or all of the “very good” or “good” genes in this study (Table 2). the effects of inclusion of the 22 tRNA-stem region sequences into the concatenated protein-coding genes were remarkable in raising bootstrap values (Fig. Evidently. 1998). see also Fig. and 7). shorter NADH dehydrogenase subunits (ND1. As expected from the higher signal/ noise ratio in the tRNA data set. 1998). such as this. the mitochondrial protein-coding genes were successively reconcatenated in order of good phylogenetic performance (Table 2) and subjected to MP analyses using the branch-and-bound algorithm with 500 replications. 9). 1998) is indispensable to resolving higher-level relationships. Bootstrap values gradually increased until genes with “medium” phylogenetic performance (7th gene) were concatenated (Fig. 22 tRNA-stem region sequences and successively concatenated NC (no 3rd) sequences were combined. no investigators have attempted to provide a guideline for choosing such a combination. despite the fact that a data set comprising all of the protein-coding genes positively misled the analyses. COIII. plus the two longer NADH dehydrogenase subunits (ND4 and ND5). best recovered the expected phylogeny. COII. COII. with all internal branches being supported by bootstrap values Ͼ90% (Fig. all terminal branches that had no close relatives were purposefully bisected by the newly determined sequences so as to avoid possible long-branch attraction. .). Boore et al. Nanba. Macey et al. Changes in bootstrap values for Clades B and C when the protein-coding genes were successively concatenated according to the ranking in Table 2. Sports and Culture. along with simultaneous attempts to find new nuclear markers.. Iwashita for their assistance in the sequencing experiments. because it is troublesome to sequence not only several protein-coding genes with good phylogenetic performance. unpublished data). such as the RAG-1 gene in birds (Groth and Barrowclough. Science. Finally. 1997) exist in complete mtDNA sequences. and graduate students at Fisheries Oceanography Laboratory. and 10660189 to M.N. 9. Further sequencing of mitochondrial genomes from deliberately chosen teleosts. Desjardins and Morais. with the exception of the data sets comprising the ND5 gene only.450 MIYA AND NISHIDA FIG. (B) amino acids (AA). 1998). ample “genomic landmarks” (Quinn. but also all 22 tRNA genes dispersed in the mitochondrial genomes. University of Tokyo for generously allowing us to use their experimental facilities. Yamamoto and K. K. excluding the entire 3rd codon positions (NC [no 3rd]). 1990. K. In fact. will undoubtedly lead to new insights into the higher-level relationships of teleosts. the same can be true for resolving the higher-level relationships of other animals. Furuya. Nishida. position TVs positively misled the analyses. A portion of this study was supported by Grants-in-Aid from the Showa-Seitoku Foundation and Ministry of Education. are prerequisites for resolving higher-level relationships of teleosts under the maximum-parsimony optimality criterion. 1999). Bootstrap values were obtained for (A) nucleotides. 09740644 and 11640705 to M.. Fukuyo. ACKNOWLEDGMENTS We thank S. A. some of which are candidates for synapomorphies for several families or lower taxonomic categories of teleosts (M. We therefore conclude that judicious choice of mitochondrial genes and appropriate weighting of the data.. in conjunction with purposeful taxonomic sampling. 1991. Miya and M.. which may serve as synapomorphies for defining monophyletic groups (e. and M. such as gene rearrangements. we have found more than 10 examples of gene rearrangements in various teleostean lineages. 1997. Note that other clades were consistently supported by bootstrap values Ͼ95%.g. Of course. but see Mindell et al. However. Pa ¨a ¨ bo et al. Kumazawa and Nishida. and (C) 22 tRNA-stem region sequences combined with successively concatenated NC (no 3rd) sequences. Japan (Nos. Iguchi for donation of a Coregonus lavaretus specimen and K. 1999) is quick and easy and allows an accurate determination of complete mtDNA sequences that is faster than sequencing cloned mtDNA (Sorenson et al.M.. 1995. 1999). Moreover. we acknowledge that the results presented here may discourage some molecular systematists. Nanba. 1995. Thanks are also due to Y. . sequencing complete mtDNA with the PCR-based approach utilized here (Miya and Nishida. Kawaguchi. Y ϭ C/T. L10267-ND3 TTT GAY 31. L3437-ND1 AAT GTK 8. H2590-16S 7.. H13396-ND5 40. L5260-ND2 CTG GST 14. L10201-ND3 TTT GAC 30. H1358-12S 5. L4546-ND1 ACK TTT 12. H14086-ND5 43. H10970-ND4 33. S ϭ G/C. H6371-CO1 21. H9375-CO1 29. H4432-Met 13. H12632-ND5 38. L2949-16S GGG ATA 6. H7480-Ser 24. L12321-Leu GGT CTT 35. 1981) by convention. L1803-16S AGT ACC 4. H5937-CO1 20. H11534-ND4M 34. H15957-Pro 49. H4983-ND2 15. H3084-16S 8. H4427-Met 12. L15765-CYB ATT CTW 45. . H3466-ND1 9. H15149-CYB 47. L12936-ND5M AAC TCM 37. L4166-ND1 CGA TAT 10. L6199-CO1 GCC TTC 19. L15777-CYB TGA ATT 46. L8984-ATP ATT GGK 26. W ϭ A/T. see Fig. H7901-CO2 25. H14080-ND5 42. corresponding to the position of the human mitochondrial genome (Anderson et al. H12145-His 36. H4866-ND2 14. H9639-CO3 30. L13731-ND5 TTA ACC 40. H11618-ND4 35. b For relative positions of primers in the mitochondrial genome. L14735-Glu AAC CAC 42. L10681-ND4 GCK TTT 32. L15411-CYB GAT AAA 44. H5334-ND2 16. H884-12S 4. H14714-Glu 45. L11895-ND4 CCT AAC 34. H8168-CO2 26. H14718-Glu 46. c Positions with mixed bases are labeled with their IUB codes: R ϭ A/G. H598-Phe 2. L12329-Leu CTC TTG 36. H10019-Gly 31. L2998-16S TCG ACG 7. L1083-12S ACA AAC 3. L5261-ND2 CWG GTT 15. L4633-ND2 CAC CAC 13. L2510-16S CGC CTG 5. L13553-ND5 AAC ACM 38. L and H denote heavy and light strands. 3. L8598-ATP TCC TGA 25. H12293-Leu 37. L3737-ND1 CAA ACW 9. L8343-Lys AGC GTT 24. H690-12S 3. H3934-ND1 11. H13069-ND5 39. L5644-Ala GCA AMT 16. L9514-CO3 TTC TGA 28. H3718-ND1 10. L7255-CO1 GAT GCC 21. H10433-Arg 32. H15973-Pro TT AAA A CCC ATG A TTA GTT AAC AGG CC T TAA WGA TWG GTG CC AAG C GC GCC CG GC GTA GA CT CTC TTG GTG CAA GT CAA GC TWG CCT G GCA AA AC CC GG GTA GA KCC G CCA ARG C AAT CCT AAG ACC T GGT TTG KCC TCA G T a Primers designated by their 3Ј ends.b Sequence (5Ј 3 3Ј) c TGC ACC ATT RGG ATG TCC TGA TCC AAC ATC TAG GCG AAC CGA GTA ACA AGA ATK ACT GCG CCG TTT AAK ATT CGK CGK AAC GGG TGG TTG CCK AGT ATG TCK CCG AAG GGG CGG CTG CAA AAC GAT GCT TGG CTA TTG GAT GTG CCT GCG AGG AGG GCG TAG TTT GGT TAG GAG TTG CAT GAG CGC CGG GCT AGT TAG GGT TCG TAT WCA AAC GGK TTW AGG AGR TGA TGC GTG ATT CCW CAG TGG GTR CAG CTT CGG ATR TGG GAC CAT TAT AGK CTG GTG CAC CAG CTG ATT ATK TAK STK GCW TTG TTG GGC GCR TTG GGA TTT GCT GGT CGG CGT GAT AAA TCT TAT TCT TGT CGW GCK CGT TAG TAG GAG CAA CCA GCC GCK CTG YTG TAT ATT AKT ATA ATG TGA KGK GGW WAG GTA ACK TTT CAA GTT GAG TTT ATG GTT AGG TTG AAT TAG KCC AAT AAG GTT CAG TGC GGC TAT YTA TGC CTG TTG GAA ACG TTG CAT AGT AGT AAG AAG TTT TGT ATG CCK GGG AAK GCT TCG TCW GTC AAK TCT GCY GTG TTT KGG ATA GAK TKG GAG ACG TGK CGG CTT TTG TAK GCK AAC TTG TCA AGG TCT AGK TGT ATG TGG ATA GTT GCT ACC ATG ATW TTG GGG GTT TTT TGK TAH TAK GWA CTT TCT AGG TCA ACT TGA TAG GAG ATG AGG CGT CAK ATT TTG GAG AWK GAG TTA TTT TAK AGT ATG CCT ATT GTT GCK AAC AAT GAA AAR CTT GGT TAW TGT CAC GGC TCG ACC TGG AAK GTT AAT TAT CGG TGT GTT AGG AGG GGA TGT TTG ATW AAG TTT AA CTT CAT GTC CTA CAT GT GGA AGC WAG GGG TAG AAC TTG AGK AGG TCT CAG AKK TTR GAG GGT WAC GGA TAT CAY RRG GCT TT A GAG GG TT AT AGT TG CC GGG GGT CA TG CT CT TCG GGT TG GA AA AC CA TG AGT AT CA AG CGA TCA TA GC TA GTT GC GC TG GC CC AT CC GGT AAC CAT CA TYT AGT S-LA-16S-H 1. L5698-Asn AGG CCT 17.MITOGENOMICS AND TELEOSTS 451 APPENDIX 1 PCR and Sequencing Primers used in the Analysis of Mitochondrial Genomes of the Three Species L primers a. H6558-CO1 22. H1903-16S 6. respectively. H9076-ATP 28.b Long PCR primers S-LA-16S-L CGA TTA PCR and sequencing primers 1. L9916-CO3 CAC CAT 29. L7863-CO2 ATA GAC 22. H8589-ATP 27. H6855-CO1 23. H15560-CYB 48. H13727-ND5 41. K ϭ G/T. H5334-ND2M 17. L6730-CO1 TAT ATA 20. H14473-ND6 44. L15927-Thr AGA GCG 47. L9220-CO3 AAC GTT 27. L5956-CO1 CAC AAA 18. L14504-ND6 GCC AAW 41. L4180-ND1 CAA CTC 11. L15998-Pro AAC TCT Sequence (5Ј 3 3Ј) c AAG TCC TAC GTG ATC TGA GTT CAG CAA TGG GCA TTT ACA AAA GTM ATT GAT ATG KCC CCW TTA TCR CAG CGA GAC CCW GGA TAC GAA GGC GGC RRC KTA TAA GCC TTT CCT CTA TCK CCW CTW AGG GTG TGG TCT GCC CWA GCT CGT TCA ATT ACM GGC TCG TAC AGC GAT AGG ACC GCG GGG GGM TCS CAA CAT AGC CGA TGC TRC ACA TCC ATT CGA ATR ACM ATT CTT CTT TGA CGA TGG TTC GGC CTR GAA GCK AAA ATG AAC CAA GAG TAY TGA TCA GCW TGT AAY YCA TGA GGC GTC CMT ATA TAG GAA AAA CAA TTT CCT TAT CTM TTR WCY GCA CMA CWA CTT TAC GGC ATA GTM CTG AAT TTA TTA CCA AAT CCC TAY TTT GGS ATY TGT GCC GGR CAA MTC ATT CTW GCC AAC GAA TAT AAC TTC ATT ATA TTG TGG GCA ATA AGC AAC TC ACG TAC GAA ATK GTW CAC GTT ART AAT TAA AAA ACC AAT TGA TGA GAC AGC AGC TG CAA ACC CA GAA TCT GC GAA CAT GAA AAA CAA CAA AAC CT GMC TAM TCA CC CAC GGM CCW TAA CTC CTG AA C TGA AA AT ACC GG GT CA TGR TG GA G GA TTA GKT CT AA GC AA CC TCG ATG TTG H primers a. H5934-CO1 19. H5669-Asn 18. L11424-ND4 TGA CTT 33. L13562-ND5M CTW AAC 39. L613-Phe TTA ACG 2. L15369-CYB ACA GGM 43. M ϭ A/C. L8329-Lys AGC GTT 23. . Roe. based on the amino acid sequence of the growth hormone. M. G. Begle. A. Classification of fishes both recent and fossil. Smith.. Evol. (1994). Ritchie. G. T. D.. L. Inst. Barrell. D. REFERENCES Anderson. A. I. H. A. J.. B.. Zool. M. Janke. J. Bernardi. Drouin. J.. T. Daehler. Evol. Sequence and organization of the human mitochondrial genome. (1991). 11: 854 – 863. Avise. L. Molecular phylogeny of bony fishes. Eperon. Battaglia. Monophyly and relationships of the argentinoid fishes. W. S. P. (1994)...” Chapman & Hall. (1993). P. Staden. Schreier. Deducing the pattern of arthropod phylogeny from mitochondrial DNA rearrangements. D. A. A. D. Coulson. Phylogenetic relationships among eutherian orders estimated . H.. Bargelloni. Trav. L. Nature 290: 457– 465.. New York. Natural History and Evolution. and Meyer. and Hasegawa. G. Sanger. Syst. G. C. Pa ¨a ¨ bo. M. Boore. B. Bankier. L. Begle. S. J. “Molecular Markers.. Biol. A. M.. Molecular evolution at subzero temperatures: Mitochondrial and nuclear phylogenies of fishes from Antarctica (suborder Notothenioidei).. and Brown. D’Onofrio. Nature 376: 163–165. and the evolution of antifreeze glycopeptides... Stanton. A.. Relationships of the osmeroid fishes and the use of reductive characters in phylogenetic analysis. M. Sci.. P. (1940). D.. R. J. and Young. Berg. Adachi... P. USSR 5: 87–517. S... L. 40: 33–53. (1994). R. I. C. T. Acad. B. S. Lambert.. and Bernardi. Mol. 37: 644 – 649.452 MIYA AND NISHIDA APPENDIX 2 Location of Features in the Mitochondrial Genomes of the Three Species Coregonus lavaretus (AB034824) a Position number Features tRNA 12S rRNA tRNA Val 16S rRNA tRNA Leu(UUR) ND1 tRNA Ile tRNA Gln tRNA Met ND2 tRNA Trp tRNA Ala tRNA Asn tRNA Cys tRNA Tyr COI tRNA Ser(UCN) tRNA Asp COII tRNA Lys ATPase 8 ATPase 6 COIII tRNA Gly ND3 tRNA Arg ND4L ND4 tRNA His tRNA Ser(AGY) tRNA Leu(CUN) ND5 ND6 tRNA Glu Cyt b tRNA Thr tRNA Pro Control region a Phe Diplophos taenia (AB034825) a Position number From To 69 1015 1086 2771 2846 3821 3896 3965 4034 5081 5155 5226 5300 5394 5464 7028 7086 7165 7869 7944 8110 8783 9568 9641 9990 10060 10357 11731 11800 11869 11943 13782 14297 14366 15512 15584 15653 16418 Size 69 946 71 1685 75 975 71 69 (L) 70 1047 71 69 (L) 73 (L) 66 (L) 71 (L) 1563 71 (L) 74 691 75 165 684 785 73 349 70 297 1381 69 67 73 1839 519 (L) 69 (L) 1141 72 70 (L) 765 Codon Start Stop Polymixia japonica (AB034826) a Position number From To 68 1019 1091 2777 2851 3826 3908 3977 4045 5092 5165 5238 5312 5416 5487 7039 7112 7188 7892 7966 8135 8808 9593 9664 10013 10083 10380 11754 11823 11892 11966 13805 14323 14392 15538 15611 15681 16481 Size 68 951 72 1686 74 975 71 70 (L) 69 1047 71 69 (L) 73 (L) 66 (L) 71 (L) 1551 71 (L) 73 691 74 168 684 785 71 349 70 297 1381 69 69 73 1839 522 (L) 69 (L) 1141 73 70 (L) 800 Codon Start Stop Codon Size 68 947 72 1679 75 975 72 71 (L) 69 1050 72 69 (L) 73 (L) 68 (L) 70 (L) 1551 71 (L) 74 691 74 168 684 785 70 349 71 297 1381 69 69 73 1839 522 (L) 68 (L) 1141 72 70 (L) 1076 Start Stop From 1 69 1016 1088 2767 2842 3824 3893 3963 4032 5083 5156 5226 5333 5401 5472 7023 7098 7186 7877 7952 8110 8793 9578 9648 9997 10068 10358 11739 11808 11878 11951 13786 14308 14380 15521 15592 15562 To 68 1015 1087 2766 2841 3816 3895 3963 4031 5081 5154 5224 5298 5400 5470 7022 7093 7171 7876 7950 8119 8792 9577 9647 9996 10067 10364 11738 11807 11876 11950 13789 14307 14375 15520 15592 15661 16737 ATG ATG GTG ATG ATG ATG ATG ATG ATG ATG ATG ATG ATG 1 70 1016 1087 2722 TAA 2847 3826 3897 3965 TAA 4035 5085 5158 5228 5329 5394 TAA 5466 7016 7092 T— 7179 7870 TAA 7946 TA– 8101 TA– 8784 9569 T— 9642 9991 TAA 10061 T— 10351 11732 11803 11871 TAA 11944 TAG 13779 14298 T— 14372 15513 15584 15654 ATG ATG GTG ATG ATG ATG ATG ATG ATG ATG ATG ATG ATG 1 69 1020 1092 2778 TAG 2852 3838 3908 3977 TAG 4046 5095 5170 5240 5351 5417 AGG 5489 7042 7116 T— 7202 7893 TAA 7968 TA– 8126 TA– 8809 9594 T— 9665 10014 TAA 10084 T— 10374 11755 11824 11894 TAG 11967 TAG 13802 14324 T— 14398 15539 15612 15682 ATG TAA ATG TAG GTG TAA ATG ATG ATG ATG ATG ATG ATG T— TAA TA– TA– T— TAA T— ATG ATG ATG TAA TAG T— DDBJ/EMBL/GenBank Accession Nos. Nierlich. Y.. Patarnello. Copeia 1992: 350 –366. J. P. F... L. (1992).. Caccio’.. Zool. de Bruijn. Collins. Mol. (1995).. (1981). H. Cao. Proc. G. M. U. Moritz. (1996). W. In “The Biology of Biodiversity” (M. Biol. Marsupialia and Eutheria. Kendall. sister group of Ostariophysi. and Wakeley. 35–52. A. 47: 3– 8. (1993).. de Pinna. Y. Gemmell. and Morais. Greenwood. and Perasso. (1984).” Freeman. and Nishida. (1996). Basal euteleosts: Relationships. M. Dowling. Sci. San Diego.. Evol. Janvier.. (1999). Nature 369: 684 – 685. Sci. V. Lee. 150: 95–197. M. Complete sequence of a sea lamprey (Petromyzon marinus) mitochondrial genome: Early establishment of the vertebrate genome organization. M. Mol. L. D. (1989). D. 2nd ed.. and Arnason. Kocher. C.). Chang. In “Ontogeny and Systematics of Fishes” (H. Phylogenet. 5: 203–214. Curole. 12: 115–123. Xu. Sunderland. T. Johnson. Urbana. C.. A. D. Pa ¨a ¨ bo. No. S. Long PCR. P. Otto. C. G. Mus. Mol... pp. Richards. P. In “Interrelationships of Fishes” (M. 42: 153–159. New York. G.. M. Mol.. L. Evol. D. N. X. J. Teleostean monophyly. Tokyo. Kumazawa. Syst. R. and Dixon. In “Interrelationships of Fishes” (M. Feldmaier-Fuchs. Johansen. Monophyly of the euteleostean clades—Neoteleostei. “SeqPup Manual Aligner for Macintosh Ver. and Gachelin. and Barrowclough. Rosen. M. Stiassny. Moser. W. Acad. Ichthyol. Kumazawa. Evol. Biol. S. T. H. (1999). Eds. In “Interrelationships of Fishes” (M. The complete mitochondrial genome of the wallaroo (Macropus robustus) and the phylogenetic relationships among Monotremata. The complete nucleotide sequence of the mitochondrial DNA of the dogfish. Trends Ecol. T. Gardiner.. A.. K. W. P. Mitochondrial molecular clocks and the origin of euteleostean biodiversity: Familial radiation of perciforms may have predated the Cretaceous/Tertiary boundary. H. D. In “Interrelationships of Fishes” (M. G. (1996). Parenti. Sinauer. P. Eds. Yamaguchi. and Pa ¨a ¨ bo.. (1996). W. D. J.” Indiana Univ. Biol. D. Meyer. J. Genetics 139: 873– 887. and Kocher. (1988). Mol. 38: 138 –155. D.. W. Relationships of lower euteleostean fishes. D. J. C. N. Hennig. Sampling properties of DNA sequence data in phylogenetic analysis. Clupeomorpha.). L. D. Parenti. K. E. J. L.-J. Mar. Stiassny. Soc. C. MA. (1993). (1966). Variations in mitochondrial tRNA gene organization of reptiles as phylogenetic markers. 38: 297–309. Genetics 150: 331–344. C. (1992).. Nat. (1966). (1990). 0. 16: 784 –792. L. 52: 554 – 626. Cummings. 212: 599 – 634. Biol. von Haesler. (1993).. pp. Ed. Hillis. 453 Hendy. Biol. (1996). L. and G. Weitzman. R. Mar.. G. Groth. Biol. Mol. Johnson. “Phylogenetic Systematics. M. San Diego. Relationships of the stomiiform fishes (Teleostei). A. Mitogenomics: Digging deeper with complete mitochondrial genome. Dynamics of mtDNA evolution in animals: Amplification and sequencing with conserved primers. 2: 31–51. (1996). C. V. Ribosomal DNA: Molecular evolution and phylogenetic inference. Mol.. (1994). F.). Y. (1989). and Ctenosquamata... Parenti. R.. P. Proc. Huang. Cheng. (1996). G. E. 150: 31–93. P.. Cao. V. R. with a provisional classification of living forms. Biol.. Hillis. M.. S. and Roe. (1983). L. Lecointre. Zool. A framework for the quantitative study of evolutionary trees. C. Eds. S. and Hasegawa. R. Comp. R. Palmer. The phylogenetic utility of the . J. Lydeard. (1994).. Spec. Stiassny. Janke. F. 333–353. Lecointre. Lawrence. Natl. Stiassny. F. C. Cohen. N. K. F. phylogenetic accuracy. San Diego. Evol. M. J. Scyliorhinus canicula. Percomorph phylogeny: A survey of acanthomorphs and a new proposal. G.. (1995). P. (1996). Desjardins. Evol. D. Fahay. Q. and G. 249 –320. D. J. J. Sci.. 12: 814 – 822.. S. Parenti. Genes and other samples of DNA sequence data for phylogenetic inference.. Eurypterygii. and Weitzman. P. Phyletic studies of teleostean fishes.. C. T. Fink. Moritz. G. (1995). Edwards. Biol. Am. M. Biol. Bull. Bull. Spruyt. (1999)... 251–332. W.6. pp. H. and Littlewood.. (1991). and S. D. Mus. S. Thomas. Interrelationships of stomiiform fishes. S. Johnson.. 131: 339 – 456. Johnson. V. Johnson.. Evol. D. S. Copeia 1992: 8 –25. 15: 1637–1646.. A. and Nishida. D... Jr. and G. Bull. C. G. Sequence evolution of mitochondrial tRNA genes and deep-branch animal phylogenetics. San Diego. (1982). Mar. A.. Carroll. Mol. Kumazawa. San Diego. Syst. G. Bull. Eds.. H.. and Respess. Complete mitochondrial DNA sequences of the green turtle and blue-tailed mole skink: Statistical evidence for archosaurian affinity of turtles. and Wilson. and Rieseberg. (1999). L. The mitochondrial genome of a monotreme—The platypus (Ornithorhynchus anatinus). 193–207. Delmarre. Evol. and Nishida. Herpetol. Phylogenet. Eds. (1998). J. USA 94: 1276 –1281. D. J. J.” Univ. Interrelationships of basal neopterygians. Okada. “Vertebrate Paleontology and Evolution. (1999). and Myers. 1. K. and Kocher. B.. 147–174... Stiassny. Johnson. Johnson. H. and investigator bias. P. (1993). pp. Y. with a description of Diplophos. W.. The complete mitochondrial DNA sequence of the shark Mustelus manazo: Evaluating rooting contradiction to living bony vertebrates.. J. Percomorph phylogeny: Progress and problems. L.). Sci. 196: 345–350. L. Evol.. L. Eds. R. and Penny. A 28S rRNA-based phylogeny of the gnathostomes: First steps in the analysis of conflict and congruence with morphologically based cladograms. (1998).. Academic Press. 43: 174 –193. Bull... Zool. pp. Y. (1995). Chang. D. S. 37: 380 –398. (1997). S. Illinois Press. Bloomington. J. Mol.. Evol.). R. H. J. Mol. Le ˆ . L. Eds. Kumazawa. In “Molecular Systematics” (D. Richardson. Laudet. and Lo. and Patterson. G. J. S. pp. Evaluating the phylogenetic utility of genes: A search for genes informative about deep divergences among vertebrates. Hist. 14: 394 – 398. Springer-Verlag. Gilbert. J. Sequence and gene organization of the chicken mitochondrial genome. Graybeal. and G. B. G. (1996). Bull. A. Fink. P. Biotechnol. Y. Otto. (1996).MITOGENOMICS AND TELEOSTS from inferred sequences of mitochondrial proteins: Instability of a tree based on a single gene. T. Acad. Academic Press. 117–146. Gravitt. and Nishida. pp. Syst. Delarbre. Nucleic acids III: Analysis of fragments and restriction sites. L. Johnson. KS. Mable. G. and Patterson. T. The evolution and interrelationships of the actinopterygian fishes. 12: 759 –772. Mol. Kato. G.. L. Barriel. Mol. M. G. D. Parenti. Academic Press. P. pp. Mus. and Liem. M. M. Gallut. Rev. M. and Bakke. Natl. J. 39: 519 –527.. L. Publ. Lauder. L. D. The complete mitochondrial DNA sequence of Atlantic cod (Gadus morhua): Relevance to taxonomic studies among codfishes. (1994). D. Janke. J. G.... Y. Hillis. Evol. and B. D. L. Academic Press. J. J.). Am. Evol. G. and Nelson.. In “Interrelationships of Fishes” (M.-Y. S. M. Maisey.. Mol. Academic Press. Waddell. V. The complete nucleotide sequence and gene organization of carp (Cyprinus carpio) mitochondrial genome. Basal divergence in birds and the phylogenetic utility of the nuclear RAG-1 gene. I.. S. S. Taxonomic sampling.. A. S. Zool. and Weitzman. and G. Cummings. M. Comp.. 66: 411– 453. 52: 3–28.. T. Johnson. USA 86: 6196 – 6200.). F.. 202–206. (1998). Villablanca.).. Harold. and Wakeley. R. Biol.. D. USA 95: 10693–10697. Wiley. 9: 278 –280. (1996). Thomas. G. G.. Mar.. Russo. In “The Hierarchy of Life” (B. M. Molecules and morphology in studies of fish evolution. Swofford. Molecular evolution of the mitochondrial genome. Nucleic acids II.. 153: 265–326. S. C. Ortı ´.. Zool.. B. Academic Press. A. M. Phylogenetic relationships of neopterygian fishes. Y.454 MIYA AND NISHIDA and the limits of resolution of mitochondrial ribosomal DNA sequences. Noack. Shortcomings of the cytochrome b gene as a molecular marker. Mus. R. Stiassny. and T. P. Additional support for Afrotheria and Paenungulata. Mommsen. (1985). New York. Molecular phylogeny and evolution of the deep-sea fish genus Sternoptyx. Mus.. Phylogenet. A. D..).. M. Elsevier. M. London. (1997). H. Academic Press. (1997). Miya. Syst. Am.. A. and Nei. J. A.” Wiley.. T. Hillis. Nelson. (1994). and Dores. Primers for a PCR-based approach to mitochondrial genome sequencing in birds and other vertebrates. Amrine. Kumazawa. T. R. 285–303. Williams. (1991). M. F. Zardoya. Miya. Organization of the mitochondrial genome of a deep-sea fish Gonostoma gracile (Teleostei: Stomiiformes): First example of transfer RNA gene rearrangements in bony fishes. M.. P. pp. Penny. M. and Arnason. C. Efficiencies of different genes and different tree-building methods in recovering a known vertebrate phylogeny. and Johnson. (1995). Sunderland. P. D. D. Multiple independent origins of mitochondrial gene order in birds. and Brown. A. J. W. (1998). (1996). Evol. and H. W.. Janke. Wilson. Rubin. Molecular phylogenetic perspective on the evolution of the deep-sea fish genus Cyclothone (Stomiiformes: Gonostomatidae). G.-S. M. Rearrangements of mitochondrial transfer RNA genes in marsupials. Fernholm. Bremer. Parenti. D. Stiassny. E. R. A review of the pelvic girdle of acanthomorph fishes. J. New York. D. Hasegawa. M. Biol. 48: 65–75. and Wong. Mol. and Baldwin.. M. Bull. 104: 209 – 242. pp. K. Academic Press. and Arnason.. G. (1984). and Meyer. Maddison. (1998). 4: 129 –138. In “Avian Molecular Evolution and Systematics” (D. Nelson. D. D. Eds. Molecular phylogeny of the deepsea fish genus Gonostoma (Teleostei: Stomiiformes): Two paraphyletic clades and resurrection of Sigmops.. Congruence between molecular and morphological phylogenies. (1974).. A. Phylogeny and zoogeography of salmoniform fishes and relationships of Lepidogalaxias salamandroides. E. J. P.. Mar. 397–513. Evol. pp. (1985).” 2nd ed. G. Interrelationships of higher teleostean fishes. (1997). R. 1–11. Mol. M. J. Evol. Inferring phylogenies from mtDNA variation: Mitochondrial-gene tree versus nuclear-gene tree. Nature 389: 803– 804. M. and Nishida. Nelson. M. Rasmussen. Natl. D. Evolution of mitochondrial DNA in fishes. M. J. Miya. The complete nucleotide sequence of the Xenopus laevis mitochondrial genome. San Diego. and Moore. W. In “Molecular Systematics” (D. C. M. San Diego. Kocher and C.. L. J. Sunderland. Pa ¨a ¨ bo. E. L. 3–28.). C.” Sinauer. S. A. A. M. Biol. 3. inferred from mitochondrial DNA sequences. and Humphries. 24: 153–188. Amsterdam. C. Yuri. N. J. Johnson. and B. Zool. L. Washington. (1993). Biol. mitochondrial cytochrome b gene for inferring relationships among actinopterygian fishes. “Fishes of the World. Rosen. Hochacka. R. A. (1994). J. In “Interrelationships of Fishes” (P. Z. and Mindell. The radiation of characiform fishes . Proc. In “Molecular Systematics of Fishes” (T.. 325–336. P. M. 13: 525–536. Evol. (1997). Biotechnol. and Nishida.1. (1999). A. G. P. Evol. D. (2000). “PAUP: Phylogenetic Analysis Using Parsimony. M. S. pp. Mindell. Bull. “Interrelationships of Fishes. and Papenfuss. Wiley. D. Mindell. Whitfield. C. DC.. D. The polymerase chain reaction. Patterson. and Maddison. Mol. M..). MA. J. S. T. Moore. Eds. L.. A. W. J. (1993). K. M.. D. D. Greenwood. D. Structural biology and phylogenetic estimation. and Meyer. (1993). and the limits of inference based on comparisons of sequences. Am. Chem. U. J. 47: 61–76. (1973). D. B. Eds. W. Ast.). Amphioxus mitochondrial DNA. Din-Pow. J. Normark. Rosen. Nelson. P. (1996).. (1976). Proc. Linn. D. “MacClade Ver. Waddell.. Mol. Rasmussen. (Eds. Olney. chordate phylogeny. Hasegawa.. Rev. Phylogeny of major fish groups. Phylogenetic. S. and Hendy.. D. Acad. 46: 75–100. pp. P. 48: 76 –93. Speciation in the open ocean. J. San Diego. Stepien. Kocher and C.. U. (1999). The mitochondrial DNA molecule of the hagfish (Myxine glutinosa) and vertebrate phylogeny. “Fishes of the World.. 2nd ed. and Stanhope. M. (1999). Naylor. K. Natl. M. and Yuri. (1999)... Sci. R. N.. Evol. Syst. Eds. Sinauer. C. 52: 137–169. C. 43: 375–398.. Mol. Biol. Ed.). M. Nat. E. (1992). Phylogeny of lampridiform fishes. Acad... M. Obtaining a more resolute teleost growth hormone phylogeny by the introduction of gaps in sequence alignment. (1997). Dimcheff. J. Syst. and Harrison. M. G. Sci.. New York. Rosen. MA. An essay on euteleostean classification. The complete mitochondrial DNA sequence of the bichir (Polypterus ornatipinnis). T. Ver. (1989). K. Ecol. A. Syst.. J. Amsterdam.” 3rd ed. Meyer... and Nishida. Syst.” Academic Press. J. Mol. the performance of mitochondrial versus nuclear genes. Molecular studies suggest that cartilaginous fishes have a terminal position in the piscine tree. (1996). 260: 9759 –9774. Fang. C. J. Genetics 144: 1165–1180... D. and Brown. Stepien. 33: 426 – 430. Patterson. Copeia 2000: 378 –389. R. 12: 105–114. (1995). Ananjeva. Stiassny. A. K. J. a basal ray-finned fish: Ancient establishment of the consensus vertebrate gene order. E. (1998). D. J. (1991). USA 96: 2177–2182. Ast. Hist. Palumbi. J. W. Burk. Evolution 49: 718 –726. 14: 91–104. B. pp. Takezaki. B. Phylogenet. Soc.... M. 2782: 1– 45. M. 1: 416 – 426. Academic Press. Interordinal relationships of birds and other reptiles based on whole mitochondrial genomes. Biol.1. R. and Nishida. Res.” Smithsonian Inst. Quinn.. pp.. San Diego. M. K.). Springer. Sorenson.. 48: 138 –152.. M. (1986). 205–221. Miles. In “Molecular Biology Frontiers: Biochemistry and Molecular Biology of Fishes” (P. Jo ¨ rnvall. Meyer. “Fishes of the World. Ichthyol.). Stepien. W. London (B) 1986: 411– 460. Biol. T. M.).. 1–38. Naylor. 10: 11–22. A. Elsevier. E. (1996). and Nishida. with comments on hypotheses of acanthomorph interrelationships. E. (1998). Trends Ecol. and C. Dimcheff. Eds. In “Molecular Systematics of Fishes” (T. 8: 819 – 834.. A. L.. Sci. J. Syst.-S. and Wilson. McCune.. Biol. The limits and relationships of the acanthomorph teleosts. S. A.. and Kocher. D. Biol. Mammalian evolution: Timing and implications from using the LogDeterminant transform for proteins of differing amino acid composition. (1999). (1997). 46: 382–388. J. Evol. Roe. Macey. D. Eds.. A. Miya. Evol. Mindell. and the impact of data partitions with heterogeneous base composition. C. S. A. H. D. Sorenson. Two novel gene orders and the role of light-strand replication in rearrangement of the vertebrate mitochondrial genome.. (1999). Biol. Mol. Sorenson.. Miya.. M.. 3. S. Annu.. Moritz. (1992). P. R. Mable. and Dimcheff. Evol. M. Nature 388: 527–528. (1993). Novit. M. R. Larson. Mol. K. Olsen. S. Phylogenetic performance of mitochondrial protein-coding genes in resolving relationships among vertebrates.. . D. Hui. Biol. P.0. Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of human and the apes. J. C. Version 4. Oncorhynchus mykiss. G. A. C. Yamaguchi. Syst. F..... L. Genetics 142: 1249 –1263. A. and Hillis. Gibson. (1983). and Meyer. Evol. D. The complete nucleotide sequence of the mitochondrial genome of the lungfish (Protopterus dolli) supports its phylogenetic position as a close relative of land vertebrates.). and Meyer. and B. Eds. J.. Shen. L.. Zardoya. 2nd ed. M. 25: 4876 – 4882. Nucleic Acids Res. Okiyama. (1981). Evolution 37: 221–244. Jeanmougin. M. (1997). Mable. Molecular phylogeny and larval morphological diversity of the lanternfish genus Hygophum (Teleostei: Myctophidae). R. 20: 4853– 4858. Evol. D. M. and Nishida.. The Clustal X windows interface: Flexible strategies for multiple sequence alignment aided by quality analysis tools.” Sinauer.. pp. P. Sunderland. C. (2000). Hauf. Hillis. including amino acid-invariant sites-LogDet plus site stripping. Tzeng. M. R. Swofford. Sunderland. A. to 455 detect internal conflicts in the data. M... The complete nucleotide sequence of the Crossostoma lacustre mitochondrial genome: Conservation and variations among vertebrates. and Hasegawa.. J. Wiley.. C. J. D. “Phylogenetics: The Theory and Practice of Phylogenetic Systematics. J. Mol. Biol. Thompson. J. Mol. Garrido-Pertierra. Zardoya. Sinauer. 15: 103–114. T. Waddell.. Phylogenetic inference. D. R. Waddell. (1995).. Evol. P. “PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods). MA. and Bautista.. Moritz. Templeton.. MA. G. (1998). F. Phylogenet. 41: 942– 951.. R. 13: 933–942. (1992). 407– 492.. Nucleic Acids Res. A. Miya. E. (1996a). The complete nucleotide sequence of the mitochondrial DNA genome of the rainbow trout.. Mol. Y.” Wiley. J. 48: 31–53. (1999). armadillo. M. Using novel phylogenetic methods to evaluate mammalian mtDNA. S. (1996b). (1996). and Higgins. with special reference to the positions of hedgehog. and Huang.. M. and elephant. Plewniak. O. In “Molecular Systematics” (D. New York. C. J. Cao.MITOGENOMICS AND TELEOSTS Swofford. M. Zardoya. F.